Introduction to incoreon AI

This page describes incoreon AI: What is the platform, how does it work, and how do the areas fit together? Every technical term is explained at its first mention; the complete glossary is part of the Resources chapter (see Glossary).

Welcome to incoreon AI

incoreon AI is a platform where AI agents take over everyday work tasks – like digital specialists, each responsible for exactly one area of work. An agent is an AI program with a defined task, its own knowledge, and its own capabilities, e.g. web search or e-mail access.

Examples from everyday work:

• Pre-checking vacation requests so that only unclear cases go to a person
• Capturing contract data from documents and providing it on request
• Automatically summarizing incoming e-mails and filing them as information

Agents deliver suggestions, drafts, and decision support. The responsibility for the result stays with the people – agent answers should be reviewed before use (see Reviewing and Refining Answers).

How incoreon AI Works

Work in incoreon AI begins in two ways: with a task in the chat or automatically via automation.

The path in the chat – five steps:

1. Set the task: A question or task is entered in the chat; files can be attached. Writing these inputs is called prompting, the input itself is called a prompt.
2. Select the agent: Before sending, the suitable agent is chosen – via the @ icon above the chat input area. Every agent brings its own knowledge and capabilities; without a selection, incoreon answers without agent knowledge – internet search can be activated via the @ icon.
3. Agent works: The selected agent understands the task, shows its reasoning steps in the thought process, and draws on its knowledge and capabilities. The knowledge comprises company-internal files and network drives stored for the agent – for example, policies, contracts, or manuals. The data stays in the company. For subtasks, the agent brings in subagents as needed – like delegating to a specialized expert.
4. Receive the result: The answer appears directly in the chat. To share a result with others, the agent can be instructed to create it as an information – it then appears in the “Information” area.
5. Continue the task: If a workflow requires a decision or processing, the agent creates a to-do. The processing runs back to the agent via the chat – the workflow then continues automatically.

The path via automation – entirely without a manual start:

An automation connects a trigger with tasks. Two triggers are available: “Inbox” starts the agent on every incoming e-mail of a mailbox, “Schedule” starts it at set times. The agent works through the tasks as workflow orchestrator: It creates to-dos and information items for the responsible persons, collects the results, and carries the workflow through to the end automatically (see Automation and Workflows).

Key Terms at a Glance

These terms appear throughout the documentation:

Term	Meaning
Agent	An AI program specialized in a specific task – like an employee who is responsible for exactly one area of work. For this, it receives a fixed instruction, matching knowledge, and capabilities (e.g. web search or e-mail access).
Chat	The central interface for exchanging with agents – tasks are entered, files are attached, and answers are received.
Prompt / prompting	The text input in the chat: a question or task for the agent. Skillfully phrasing this input is called prompting.
Thought process	The steps of thought the agent goes through before its answer; they are shown in the answer in a dashed speech bubble. Technically also called reasoning.
Language model	The software inside an agent. It understands text inputs and generates answers.
Instruction	The permanently stored base description of an agent: Who is it, what should it do, in what frame? It applies to every conversation.
Capabilities	Functions an agent can use, e.g. web search, calculations, or reading a mailbox.
Subagent	An agent that another agent brings in for a subtask.
Token	Building blocks into which a language model splits text – roughly a word or part of a word. The number influences how much text can be processed, and the costs.
Context window	The agent's short-term memory: everything said or uploaded in the current conversation. It is limited.

Full definitions are in the glossary (see Glossary).

The Platform Areas at a Glance

Area	What the area is for
Chat	Delegating tasks, having files analyzed, and asking questions – the central interaction surface.
Information	Receiving short messages and updates: results from agents, reports from automations, and messages from colleagues.
To-Dos	Working through tasks and decisions – collected centrally, with deadlines and assignments.
Understanding Agents	Finding the right agent: own and shared agents – Customized Agents on request.
Automation and Workflows	Running workflows without a manual start – via e-mail trigger or schedule.
Profile	Managing personal settings and connecting own accounts.

First Steps in Practice

The quickest way to a first result leads through the chat:

1. Open incoreon AI in the browser and sign in with the personal account.
2. Start a new chat.
3. Select a suitable agent – via the @ icon above the chat input area. Every agent brings its own knowledge and capabilities; without a selection, incoreon answers without agent knowledge – internet search can be activated via the @ icon.
4. Ask a question or formulate a task, e.g. “Summarize the key points of the attached document.” To do so, upload a file if needed.
5. Read, review, and reuse the answer – the result can be copied or downloaded as a file.

All further steps – uploading files, creating to-dos – are described in the Chat chapter.

FAQs about the introduction

Do I need prior knowledge of AI to use incoreon AI?

No. Anyone who can write e-mails and manage files on a computer can work with incoreon AI. Technical terms are explained at their first mention in this documentation (see Key Terms at a Glance).

How do I get access to incoreon AI?

Access is granted by the company's administration. If access is missing, the responsible administration helps (see Roles and Permissions).

Where do I get help if something does not work?

First in the FAQ sections of the chapters and in the Troubleshooting chapter (see Troubleshooting and Support). If the problem cannot be solved, the chapter also describes the way to support.